Hi Rich and Marco,
the most important line is about 15 lines above:
CI = 1.6449; % = spm_invNcdf(1 - 0.05);
has to be replaced by
CI = 1.9600; % = spm_invNcdf(1 - 0.025);
Therefore you have to multiply contrast.interval with
1.9600/1.6449 = 1.1916.
Hope this helps,
Michael.
Am 26.03.2010 09:03, schrieb Marko Wilke:
> Hi Rich,
>
>> I am interested in interpreting the parameter estimates provided by
>> SPM5. I notice SPM5 provides estimates with 90% confidence intervals.
>> Is there a reason 90% was chosen instead of the more standard 95%? Is
>> it possible to change the interval to 95%?
>
> Don't know about the reason, but what you are looking for is done in the
> lines following 298 in spm_graph:
>
> % compute contrast of parameter estimates and 90% C.I.
> %--------------------------------------------------------
> cbeta = SPM.xCon(Ic).c'*beta;
> SE = sqrt(diag(SPM.xCon(Ic).c'*Bcov*SPM.xCon(Ic).c));
> CI = CI*SE;
>
> contrast.contrast = cbeta;
> contrast.standarderror = SE;
> contrast.interval = 2*CI;
> assignin('base','contrast',contrast)
>
> ... and so on. Note that the results are stored in a variable called
> contrast, so even without having the bad feeling ot fiddling with the
> code ("stick to the code", as the pirates say :), you can always use
> this variable and plot it as you wish.
>
> Hope this helps,
> Marko
>
> PS: I just saw you mentioned spm5, while my lines are from spm8's
> spm_graph, but I think the same applies.
>
--
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
Dr. Michael Erb
Sektion f. experimentelle Kernspinresonanz des ZNS
Abteilung Neuroradiologie, Universitaetsklinikum
Hoppe-Seyler_Str. 3, D-72076 Tuebingen
Tel.: +49(0)7071/2987753 priv. +49(0)7071/61559
Fax.: +49(0)7071/294371
e-mail: <[log in to unmask]>
www: http://www.medizin.uni-tuebingen.de/nrad/sektion/
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
|